International Journal of Environmental Science and Development

Open access



Volume 15 Number 3 (2024)

Home > Articles > All Issues > 2024 > Volume 15 Number 3 (2024) >
IJESD 2024 Vol.15(3): 157-168
doi: 10.18178/ijesd.2024.15.3.1481

Comparison Between Machine Learning Techniques for Early Air-Pollution Detection: A Case of Bojanala Platinum District Municipality (BPDM), South Africa

Koyana.Ntombikayise1,*, Elisha D. Markus1, Malusi Sibiya2, and Adnan M. Abu-Mahfouz3
1Department of Electrical, Electronic and Computer Engineering, Central University of Technology Bloemfontein, South Africa
2Department of Computer Science, University of South Africa, Johannesburg, Florida campus
3Council for Scientific and Industrial Research (CSIR), Pretoria, South Africa
Email: (K.N.); (E.D.M.); (M.S.); (A.M.A-M.)
*Corresponding author
Manuscript received July 7, 2023; revised August 22, 2023; accepted November 9, 2023; published June 24, 2024

Abstract—Air pollution has been one of the major threats to modern livelihoods. It is affecting, health, economies, and social well-being and has even resulted in fatalities in certain instances. As part of the fourth industrial revolution in South Africa, there has been a recent focus on smart cities. Hence, exploring new ways of combating the menace of air pollution has become pertinent. Machine learning techniques have been applied to solving many modern problems. However, in South Africa, these technological solutions are still at their inception. This paper proposes an early air pollution detection technique for a city in South Africa. This city has experienced air pollution problems in the past owing to the presence of many mining and industrial activities. Past data collected in the city shows a pattern of air pollution threatening the city’s fabric. If left unchecked, the result on health and livelihoods would be disastrous. Deep learning neural networks, multiple linear regression, and random forest trees for regression are used to model the pollution patterns, and a short-term prediction strategy was designed to warn residents ahead of impending catastrophes that could be damaging to their health. Based on the results, the random forest regressor model provided better predictions and is recommended for deployment. The results therefore show that the early air pollution detection strategy provides good outcomes and could effectively monitor and warn residents and authorities.

Keywords—machine learning, air pollution, PM2.5, forecasting, regression model


Cite: Koyana.Ntombikayise, Elisha D. Markus, Malusi Sibiya, and Adnan M. Abu-Mahfouz, "Comparison Between Machine Learning Techniques for Early Air-Pollution Detection: A Case of Bojanala Platinum District Municipality (BPDM), South Africa," International Journal of Environmental Science and Development vol. 15, no. 3, pp. 157-168, 2024.

Copyright © 2024 by the authors. This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited (CC BY 4.0).